onsemi: A Research Report Through the Lens of a Long-Term Business Owner

Hunting ten-year five-baggers among great growth stocks — pressing the upside question: "Can it get much bigger?"

• QHow high is its market ceiling? Is it growing a slice of an existing pie, or creating an entirely new market?5/10

  A

  The ceiling is high, but onsemi is mainly "growing an existing large pie" rather than conjuring an entirely new market out of nothing — its growth comes from rising penetration of "stuffing more power/sensing chips into each system" across the three existing arenas of electrification, industrial automation, and AI-data-center power, not from opening up new zero-to-one demand.

  First, the arena's absolute ceiling. Per the report citing WSTS figures, the global semiconductor market will approach $975 billion in 2026, but the main engine driving that growth is memory and logic, while the power/analog/industrial demand more relevant to onsemi recovers more slowly. The sub-segment where onsemi truly holds position — silicon carbide (SiC) — does show structural high growth: on third-party figures, the SiC market is expected to grow from about $3.8 billion in 2025 to about $12 billion in 2030, a CAGR in the low-to-mid 20s percent. This is a pie that is genuinely getting bigger, and onsemi is indeed one of the world's top five players.

  But to be honest: this is "growing an existing pie," not "creating a new market." EVs, 800V drivetrains, industrial variable-frequency drives, and AI-server power are all already-existing markets with clear customers and applications; what onsemi does is push up its share and the silicon content per unit. The report is very clear about the company's core positioning — it is "an important participant that holds a real position in high-growth arenas yet still sits in a landscape surrounded by strong rivals," with rivals Infineon, ST, NXP, and TI long on the scene. This is fundamentally different from what Baillie Gifford's LTGG most favors — "creating a market that did not previously exist" (like early EVs or streaming).

  Down to its own numbers, onsemi's full-year 2025 revenue was $5.995 billion, down 15% year-over-year, and even in the peak years of 2022-2023 it only reached just over $8 billion — relative to the total pool of its arena, the penetration room is indeed still large, but share gains have to come from head-on fighting for existing volume, not from enjoying an uncontested new blue ocean.

  Ceiling verdict: the slope is real, but the snow is neither thick enough nor exclusive enough. The market ceiling is not a constraint for onsemi; the constraint is "how much it can carve off this existing large pie, and at what margin" — a question of penetration and competitive landscape, not of market size. Treating it as a "disruptor opening a new market" would overstate the nature of its growth; treating it as a "strong contender fighting for share in structurally expanding power semiconductors" is accurate.

  Jun 11, 2026
• QOver the next five years, can its revenue at least double? Is growth driven mainly by volume, price, or new businesses?4/10

  A

  Bottom line up front: onsemi "doubling" revenue over the next five years (from about $6 billion in 2025 to $12 billion) is a long shot; the more realistic script is to first repair back to the prior high of just over $8 billion in 2023, then grow moderately — unless AI-data-center power + SiC drivetrains see a boom far beyond the current pace. Even if it does grow, the driver is mainly "volume" (shipment recovery + higher silicon content per system + new-application penetration), while "price" is a headwind in a bad cycle and only turns into a tailwind in a good one.

  First, the starting point. 2025 was a cyclical trough, with revenue of $5.995 billion, down 15% year-over-year, and all three segments declining across the board: PSG $2.805 billion (-16%), AMG $2.262 billion (-13%), ISG $928 million (-17%). "Doubling" off this low base looks not hard, but onsemi's all-time high was only $8.3 billion in 2022 — doubling in five years would mean pushing the historical peak up another 45%, which for a highly cyclical power-semiconductor company that fights head-on for share is an aggressive assumption, not a base-case one.

  Now the 2026 inflection signals — the driver is indeed on the "volume" side. Q1 2026 revenue was $1.513 billion, up 5% year-over-year, the first return to growth after seven quarters of decline; within it, automotive revenue was up 5% year-over-year, and AI-data-center revenue more than doubled year-over-year, up about 30% sequentially. Q2 2026 guidance is for revenue of $1.535-1.635 billion. Annualized, this is the slope of climbing up from a trough, not a doubling-style exponential curve. AI data center is the fastest increment right now (the Qorvo SiC JFET acquisition is aimed precisely at server AC-DC power), but it is doubling off a very small base, so its pull on the $6 billion pool still needs a few years of accumulation to become significant.

  Breaking down "volume / price / new businesses":

  • Volume: the main engine. Electrification (rising SiC/power content per vehicle), industrial VFDs, and AI-server power all follow the "stuff more chips into each system" logic — a structural tailwind.
  • Price: double-edged. In 2025 PSG gross margin slammed from 41.3% to 24.5%, precisely the result of a triple hit from volume + price + utilization (see the report); ASP is a headwind during loose supply and only turns positive once the cycle recovers.
  • New businesses: AI-data-center power + SiC JFET are real increments, but their share is still small, and the report also flags that the company's segment-level disclosure is limited, so scaled delivery is in doubt.

  Honest verdict: doubling in five years belongs to the "optimistic scenario," not the "reasonable scenario." The report's own neutral DCF assumes only 10% growth in the first five years and 5% in the next five — about 60% cumulative growth over five years, falling short of doubling. Treating "doubling" as an investment premise would be overly optimistic; a steadier expectation is "cyclical repair + single-digit to low-double-digit annual growth, driven by volume."

  Jun 11, 2026
• QFive years from now, what will take the baton as the next growth engine? Does this "second curve" exist today?4/10

  A

  Bottom line up front: onsemi's "second curve" does already exist today and is already putting up numbers — namely AI-data-center power (layered with SiC use in server AC-DC / high-voltage DC power distribution). But it is still "a high growth rate off a small base," several years of scaling away from taking the baton as the main engine; the company's core base (automotive + industrial power semiconductors) remains the absolute majority, so the second curve looks more like "ignited, but not yet dominant."

  First, the current main engine: automotive + industrial power/analog/sensing. In 2025 PSG (power) was $2.805 billion, about 47% of the total — the company's lifeblood, but itself a cyclical-repair story, not a new curve.

  Evidence for the second curve — it is not a slide deck but booked revenue: in Q1 2026 AI-data-center revenue more than doubled year-over-year, up about 30% sequentially. This is the fastest-growing piece right now, and the logic is solid: AI-server rack power density is exploding, and every stage of power conversion from the grid to the chip is migrating toward SiC / high-efficiency power devices — this power-distribution chain is exactly what onsemi wants to capture. To reinforce this capability, in 2025 the company acquired Qorvo's SiC JFET (United SiC) business, aimed squarely at data-center AC-DC power applications — an act of "actively buying ammunition for the second curve," also clearly noted in the report.

  But to be honest, three buckets of cold water: First, the base is too small. "Doubling" is pretty, but if the starting point is only a single-digit share of total revenue, its pull on the $6 billion pool needs several consecutive years of accumulation to become significant; today it still cannot support the company's valuation. Second, disclosure is opaque. In "Open questions and limitations" the report explicitly notes that the segment-level breakdown of SiC, GaN, and AI-data-center revenue shares remains limited in the company's public disclosure — it is hard for outsiders to independently verify how big it really is and how high its margins are, creating the risk of a "story-telling premium." Third, it is surrounded by strong rivals. On the same AI-power pie, Infineon, ST, and TI are all fighting; onsemi is not the sole holder.

  Baillie Gifford-lens verdict: the second curve passes on "existence" (numbers are in, backed by an acquisition, the arena's direction is right), but fails on "certainty of taking the baton" — to become the main engine in five years, it must simultaneously satisfy "grow the base + hold up margins + win the share," and right now we see only the direction, not yet the evidence of scaling. The report's judgment on this is incisive: these high-value growth points must genuinely "translate into higher segment profit, not merely story-telling." Treating the second curve as already-delivered growth would be overly optimistic; treating it as "the delivery-type option most worth watching" is accurate.

  Jun 11, 2026
• QWhat is its core competitive advantage? Will this moat widen or narrow over the next three to five years?5/10

  A

  Bottom line up front: onsemi's core competitive advantage is a four-piece set — "automotive-/industrial-grade reliability certification + long product life cycles + customer design-in and re-qualification costs + vertically integrated SiC manufacturing from boule to device" — a real but medium-width moat. Over the next three to five years it is more likely to "widen slightly" (through SiC integration and new data-center certifications deepening the lock-in), but by no means widen to the point that "others cannot get in," because same-arena rivals Infineon, ST, TI, and NXP are all peers of the same magnitude; the moat shows up more as "earn more in good cycles, not get killed in bad ones" than as "earn steadily across the cycle."

  The substance of the moat, item by item:

  • Certification and switching costs (medium): once automotive-drivetrain and safety-related chips are designed in, re-qualification costs are high and platform life cycles are long, yielding repeat revenue after design-in. But this is not SaaS-style high stickiness — the report explicitly notes that standard-product orders can generally be canceled 45-120 days before shipment, so the stickiness is "structural bargaining" rather than "absolute lock-in."
  • SiC vertical integration (potential cost advantage): onsemi owns its New Hampshire SiC boules and Czech/Korean SiC wafers, and in 2025 further acquired Qorvo SiC JFET to reinforce data-center power, giving it more supply control than a pure design company — this is a relatively wide segment.
  • Scale and channel (medium): its global front-/back-end network is broad, but not overwhelming relative to Infineon/TI.

  Hard evidence that the moat "gets offset by the cycle": in the weak-demand year of 2025, the company was forced into a manufacturing restructuring, recognizing $666.9 million of net restructuring/impairment/other charges for the year, with PSG gross margin slamming from 41.3% to 24.5% — if the moat were truly wide, low utilization would not punch through profit this hard. This shows its cost advantage gets bitten back by capacity utilization when the cycle turns down; it is not an unconditional moat.

  Wider or narrower over the next three to five years? My judgment is "slightly wider, but very dependent on execution rather than natural monopoly":

  • Forces widening it: SiC integration spreading costs as capacity ramps, AI-data-center power establishing a new batch of customer-certification lock-ins, and long-platform design wins accumulating.
  • Forces suppressing the widening: share comes from fighting head-on for existing volume (per Infineon citing third-party figures, onsemi held about 8.5% of the global power discretes & modules market in 2024 — an important player, not a leader), and GaN, advanced silicon solutions, and shifts in customer roadmaps could all erode SiC's returns.

  Baillie Gifford-lens verdict: this is a medium moat that "has thresholds, can be widened through execution, but has its ceiling pressed down by strong competition." It does not clear Baillie Gifford's top tier of "getting stronger no matter the environment," but it is far more solid than companies that "only tell stories and have no barriers." Over three to five years it will most likely widen modestly, provided management executes on SiC and data centers — and that is precisely a variable, not a given.

  Jun 11, 2026
• QIf its core business were disrupted, does it have the genes to reinvent itself? How does it treat mistakes and bad news?4/10

  A

  Bottom line up front: onsemi shows the maturity of "treating mistakes and bad news quite realistically, without covering up" — facing collapsing demand in 2025, it chose to recognize the problems all at once (large impairment + manufacturing restructuring + the CEO voluntarily forfeiting his short-term bonus), a good sign. But as for "the genes to thoroughly reinvent itself when the core business is disrupted," the evidence is on the weak side: over the past decade its transformation has been "portfolio optimization and process upgrades within the main line of power semiconductors (silicon→SiC)," an extension along existing capabilities rather than proof that it can jump out of its original arena and start over. So the more accurate phrasing is "a strong error-correction culture, but self-reinvention genes not yet truly stress-tested."

  First, the strengths — how it treats mistakes and bad news, with solid evidence:

  • Booking losses all at once, not burying landmines. In 2025 the company recognized $666.9 million in total net restructuring/impairment/other charges, and in Q1 2026 it further took $329.3 million of restructuring/asset-impairment/other charges, preferring an ugly balance sheet to squeeze out the problems of excess capacity and outdated product roadmaps — this is "realism" rather than "keeping the problems hidden in inventory and fixed assets."
  • Executives sharing the pain, not passing the buck. The report notes that in 2025, because of the company's restructuring, the CEO voluntarily forfeited his short-term cash incentive (STI) — with management docking itself first when results are poor, a posture of acknowledging bad news rather than glossing over it.

  But on the item of "the genes to reinvent itself after disruption," we must honestly put a question mark:

  • Its "transformation" is extension, not a leap. Moving from traditional power devices to SiC and to AI-data-center power still relies on migrating the same existing capability set of "power electronics + manufacturing + automotive-grade certification"; there has never been a survival-level test of "the main channel being technologically replaced, forcing a wholesale switch of arena to survive." In other words, it has proven it "can optimize its core business," but not that "its core business can be overturned and still be reborn."
  • The real disruption risk comes from technology substitution. The report explicitly lists technology substitution as an important risk — SiC has opportunity, but GaN, advanced silicon solutions, and shifts in customer roadmaps could all depress returns. If a roadmap-level disruption emerges in next-generation materials/architectures, whether onsemi has the kind of genes to "cut its own losses and reinvent itself" has no historical sample to prove for now.

  Baillie Gifford-lens verdict (the chained question supplements the implicit premise "self-reinvention genes"): error-correction and candor — pass, with solid evidence; self-reinvention — in doubt, not yet stress-tested. This company is like "an honest, mistake-admitting top student with execution online," but not the kind of antifragile company that "can grow a new species out of the rubble after the table is flipped." For a long-term owner, this means you can trust it to clean up the current cycle's mess, but you cannot blindly assume it will emerge unscathed from a disruptive technology turnover.

  Jun 11, 2026
• QDoes management (especially the founder) have long-term vision, with interests deeply tied to the company? Are they willing to sacrifice current profit for five to ten years out?4/10

  A

  Bottom line up front: onsemi's management has "broadly positive long-term vision and honesty, and is indeed willing to sacrifice current profit for the long term," but the item of "founder-style deep interest alignment" does not hold — this is a company governed by professional managers; CEO Hassane El-Khoury is not a founder and his personal stake is less than 1%, a clear gap from what Baillie Gifford most cherishes: "a founder heavily invested, betting their net worth on the company." So the more accurate characterization is "above-average alignment achieved through mechanisms (compensation structure + ownership requirements + behavioral constraints), rather than deep alignment driven by emotion and net worth."

  First, "willing to sacrifice current profit for the long term" — the evidence is fairly solid:

  • In 2025 it did a manufacturing restructuring even at the cost of an ugly balance sheet, recognizing $666.9 million of net restructuring/impairment/other charges for the year and squeezing out excess capacity and outdated product roadmaps — sacrificing current profit for long-term structural health.
  • With results poor, the CEO voluntarily forfeited his short-term cash incentive (STI) — management docking itself first, posture online.

  Next, the mechanism layer of "interest alignment" — positive but not top-grade:

  • Long-term compensation: the report notes that the vast majority of the CEO's 2025 target total compensation is variable pay and equity incentives.
  • Ownership requirement: the company sets the CEO a stock-ownership requirement of at least 6x annual salary, and all NEOs meet it.
  • Behavioral constraints: executives and directors are prohibited from hedging, pledging, or short-selling company stock — precluding "talking bullish while quietly hedging."

  But two reservations must be stated honestly: First, a non-founder, with alignment that is not "bone-deep." The report discloses the CEO holds about 710,000 shares — not low in value, but the ownership stake is still less than 1%. What Baillie Gifford values most is "a founder betting their life's net worth"; onsemi here is a professional manager's "professional-grade" alignment, not "net-worth-grade" alignment. Second, capital-allocation discipline at high prices is in doubt. Management's large low-price buybacks in 2025 are a plus, but in May 2026, using a closing price of about $105.77 as the benchmark, it issued $1.3 billion of 0% convertible notes and used about $331.9 million to repurchase about 3.1 million shares, and in November 2025 it newly authorized $6 billion of buyback capacity — continuing large buybacks at a clearly not-cheap price is not necessarily a signal of "extremely rational bottom-fishing" for new investors, showing management does not always act only when clearly undervalued.

  Baillie Gifford-lens verdict: long-term vision and honesty — pass; sacrificing the present for the long term — pass, with hard-cash 2025 evidence; founder-style deep interest alignment — fail, a structural shortcoming (non-founder + stake <1%). The conclusion is "worth trusting, but not to the point of blind entrustment" — more reliable than management that "only tells stories," yet do not pay for this trust the high valuation premium that only a founder-type company deserves.

  Jun 11, 2026
• QIf it disappeared tomorrow, how much would customers miss it? Is its growth model sustainable and free of harming society or regulatory arbitrage?5/10

  A

  Bottom line up front: on "indispensability," if onsemi disappeared tomorrow, customers would suffer considerably but would not be paralyzed — in automotive-grade power devices, SiC, and automotive image sensing it is a key supplier that is already designed in and carries high re-qualification costs (short-term switching is very painful), but it is not the sole source, and Infineon, ST, NXP, and TI could fill in over the medium-to-long term. On "sustainability," its growth model is highly healthy and does not rely on harming society or regulatory arbitrage; on the contrary, it steps on directions that regulation and policy positively encourage — electrification, energy efficiency, and industrial energy saving — on this item it is almost full marks.

  First, "how much would customers miss it" (indispensability) — medium-to-strong, but not the sole source:

  • Design-in means lock-in. Once automotive-drivetrain and safety-related chips are on the vehicle, the re-qualification cycle is long and costly, and switching suppliers in the short term is extremely painful — this constitutes genuine "missing." The report also corroborates its order stickiness: standard-product orders can generally be canceled 45-120 days before shipment, and canceling custom products also requires compensating costs already incurred — switching has real friction.
  • But there are real substitutes. The report repeatedly stresses it is "not an irreplaceable monopolist"; Infineon/ST are stronger in power devices, and NXP is stronger in automotive; per Infineon citing third-party figures, onsemi held only about 8.5% of the global power discretes & modules market in 2024. That is, the "missing" is at the level of "finding a substitute costs migration effort," not "the whole industry halts without it." On customer concentration there is also a reverse caution: one distributor contributed about 11% of revenue in 2025 — its reliance on a single channel is, in a sense, more worth watching than the channel's reliance on it.

  Next, "whether growth is sustainable and free of harming society or regulation" (the second layer added by the chained question) — this item is very clean:

  • The direction moves with regulation, not arbitrage. onsemi's demand comes from electrification, power conversion, fast charging, energy storage, and industrial automation — per the report citing the IEA, 2025 global EV sales already exceeded 20 million units, accounting for 25% of new cars. It sells chips that "make systems more efficient and more power-saving," essentially pushed along by countries' energy-efficiency and carbon-emission regulations, rather than profiting from regulatory loopholes or harming consumers.
  • Governance stays inside the lines. The report found no obvious red flags of financial fraud or aggressive accounting, executives are barred from hedging/pledging/short-selling, and in 2025 it recognized the problems all at once even at the cost of a large impairment — growth quality is "genuinely selling products," not "financial engineering masquerading as growth."

  Baillie Gifford-lens verdict (the chained question's dual premise): indispensability — medium-to-strong, but replaceable, passing the "would be much missed" test but not the "cannot do without it" test; social and regulatory sustainability — high marks, growth riding a policy tailwind, governance clean. On balance, onsemi is a company that customers "cannot leave without paying a cost, yet is beyond reproach at the societal level." Its fragility lies not in "whether it does evil" but in "whether it is the only one" — and it is not the only one, which is just another way of stating that the moat's width is medium.

  Jun 11, 2026
• QWhat are this business's unit economics (gross margin, incremental returns)? Does it get better or worse as it scales? Where does the money it earns go?5/10

  A

  Bottom line up front: onsemi's unit economics are a textbook case of "high elasticity, strong cyclicality" — when the cycle turns up, getting bigger makes it markedly better (fixed costs spread out, gross margin and incremental returns jump quickly), and when the cycle turns down, a collapse in scale utilization makes it worse fast. It is not the software kind of "bigger is more profitable, smooth compounding" business, but a heavy-asset IDM where "get the cycle right and you reap a harvest, get the cycle wrong and you get bitten back." The money it earns goes mainly to two places: maintaining/upgrading manufacturing capacity (including SiC) and large stock buybacks (the company has never paid a common-stock dividend since its IPO).

  First, the elasticity of the unit economics, with very clear evidence:

  • The upward slope in a good cycle: in 2021-2022, gross margin rose from 40.3% to 49.0% and operating margin from 19.1% to 28.3% (see the report's key financial table) — this is the leverage effect of "getting bigger, spreading fixed costs, and dropping a large share of incremental revenue into profit."
  • The downward slope in a bad cycle: in 2025 gross margin slammed back to 33.1%, GAAP operating margin was just 1.4%, and net income to parent was only $121 million. The most striking is at the segment level — PSG gross margin slammed straight from 41.3% to 24.5% within a year (per the report), because of a triple hit from falling volumes + low plant utilization + unfavorable product mix. On the same production line, once utilization changes, the unit economics are two different worlds.
  • Margins are warming up: in Q1 2026 GAAP gross margin had repaired to 38.5% and non-GAAP operating margin to 19.1% — confirming that "utilization comes back, and the unit economics come back with it."

  "Better or worse as it scales?" — the answer is "it depends which side of the cycle you're on," which is exactly where it differs from a Baillie Gifford ideal name. The ideal growth stock is one where "scale itself brings structural improvement"; for onsemi it is "scale × utilization" that decides good or bad, so scale is not a monotonically positive variable. SiC vertical integration can in theory spread costs as capacity ramps and lift the unit economics, but the facts of 2025 prove it gets bitten back by utilization in a trough — so "getting better" is conditional, not inevitable.

  "Where does the money go?" — two destinations, with discipline that has both pluses and controversy:

  • Manufacturing capex: maintaining and upgrading front-/back-end capacity and SiC boule/wafer integration. Note that capex swings widely with the cycle — cash PP&E spend reached $1.539 billion at the 2023 expansion peak and was squeezed to $341 million at the 2025 trough (report table), a heavy-asset investment of "betting on the cycle."
  • Stock buybacks (the main means of return): in 2025 full-year free cash flow of $1.419 billion was essentially all used for buybacks, with shares outstanding shrinking to about 397 million by year-end 2025; in November 2025 it newly authorized $6 billion of buybacks. Low-price buybacks genuinely thicken per-share value (a plus), but in 2026 continuing to pair buybacks with convertibles at about the $105.77 level is not necessarily "acting when clearly undervalued" (a point of controversy).

  Baillie Gifford-lens verdict: unit-economics quality is "medium, highly elastic, non-monotonic." Its incremental returns are sexy in good times and ugly in troughs; scale magnifies volatility, not just profit. Treating it as a "bigger-is-steadier compounding machine" would be a misjudgment; treating it as an elastic asset where "get the cycle right and the unit economics repair violently, get the cycle wrong and utilization bites back" is accurate — and the money mainly flows back to shareholders (buybacks) and capacity (capex), a rational direction, with discipline at high prices yet to be seen.

  Jun 11, 2026
• QFor it to rise fivefold in ten years, which conditions must simultaneously hold? Are these conditions realistic? What expectations does today's share price imply?3/10

  A

  Bottom line up front: for onsemi to rise fivefold in ten years, "a large and sustained lift in the earnings center of gravity + no contraction in the valuation multiple" must hold simultaneously — starting from the current roughly $110, market cap about $43.2 billion (closing at $110.17 on June 10, 2026, market cap about $43.18 billion, already pulled back from the all-time high of $134.92 on June 3), fivefold means the market cap must reach about $216 billion, approaching the magnitude of today's Texas Instruments (about $277.9 billion). For a highly cyclical power-semiconductor company that fights head-on for share and whose current GAAP earnings are extremely weak, this is an aggressive assumption, not a base case. What today's share price implies is not "buying cash flow at a discount" but "paying in advance for a strong recovery + AI-power delivery + high-quality capital returns."

  Conditions that must simultaneously hold for a ten-year fivefold (none can be missing):

  1. A structural step-up in the earnings center of gravity, not just cyclical repair. In 2025 GAAP operating margin was only 1.4%, net income to parent $121 million; for fivefold, it must return to and stably exceed the 2024 level and no longer be punched through by the next cycle — i.e., turn "understandable but highly volatile" into "volatility converging, center of gravity rising."
  2. AI data center + SiC truly scaling into a high-margin segment. In Q1 2026 AI-data-center revenue more than doubled year-over-year, up about 30% sequentially, but the base is small and disclosure opaque; to support fivefold, it must grow from a "high-growth small business" into a "large enough and high-gross-margin profit pillar."
  3. Revenue roughly doubling with margin expansion in parallel. The report's neutral DCF assumes 10% growth in the first five years and 5% in the next five — on this, ten-year revenue grows about 1.6x, far short of "volume-and-price rising together to support a 5x market cap," so fivefold must layer on significant margin expansion, not revenue alone.
  4. The valuation multiple not re-rating downward. This is the most fragile link — see below.

  Realism verdict: of the four above, "each is individually possible, but all four holding at once leans optimistic." The report classifies this precisely as the "optimistic scenario": its optimistic DCF (starting Owner Earnings $1.5 billion, growth 12%/6%) corresponds to an intrinsic value of only $110-135/share — that is, the current price is already standing inside the optimistic-scenario value range, and a ten-year fivefold requires the company to keep outrunning this already-optimistic assumption. This is not impossible, but it needs nearly everything to go right.

  What expectations does today's share price imply (the key sub-question added by the chained question):

  • A strong recovery is already priced in. The report calculates that the current market cap corresponds to conservative Owner Earnings of about $900 million, i.e., about 55x — trailing P/E has reached about 78x, and the free-cash-flow yield is only a bit over 2%, clearly below the roughly 4.56% risk-free yield of the 10-year Treasury. The market is paying for "future profit repair + multiple maintenance," not for "current cash flow being cheap."
  • Insufficient margin of safety. The report's fair intrinsic value is $75-95 and ideal buy $70-90 — the current roughly $110 is still above the upper end of the fair range, hardly a discount.

  Baillie Gifford-lens verdict: the conditions for a ten-year fivefold are "imaginable, but need a quadruple of positives to deliver at once with no multiple contraction" — a low-probability blue-sky scenario rather than a reliable base case. Today's price already implies the optimistic combination of "recovery + AI power gaining momentum + excellent capital returns," leaving a very thin cushion for the upside; if any one condition falls through or the multiple reverts, the fivefold narrative would collapse ahead of the fundamentals. The honest conclusion: this is a position of "betting against the risk of recovery falling short, rather than picking up a panic-priced bargain."

  Jun 11, 2026
• QWhy hasn't the market realized all this yet? Is it that it can't understand it, looks down on it, or can't see far enough? What would become the "narrative inflection point"?3/10

  A

  Bottom line up front: for a name like onsemi, Baillie Gifford's line "why hasn't the market realized all this yet" actually needs to be asked in reverse — because the market not only realized it but may have already priced it beyond fair value. It is not a dusty stock buried by "can't-understand/look-down-on/can't-see-far," but a recovery + AI-power concept stock fully (even excessively) chased in the 2026 bull market: the share price rose more than 100% in a year, just set an all-time high of $134.92 on June 3, has pulled back to about $110, and has a trailing P/E of about 78x. So the real perception gap is not "undervalued and overlooked" but "whether the recovery and AI story can deliver up to the current valuation."

  Item-by-item deconstruction of Baillie Gifford's three "market hasn't seen it" types:

  • Can't understand? Doesn't hold. The report gives this company's business a comprehensibility of 4/5 — selling high-voltage/high-efficiency power and sensing chips to automotive, industrial, and AI data centers; the logic is clear, sell-side coverage is ample, and it is not obscured by complexity.
  • Look down on it? Also doesn't hold. The market in fact thinks quite highly of it: in Q1 2026 revenue was up 5% year-over-year, the first return to growth in seven quarters, and AI-data-center revenue more than doubled year-over-year — these inflection signals have been warmly embraced by the market and driven the share price to new highs, hardly disdained.
  • Can't see far? This is the only two-way perception gap that could exist. The bulls' "seeing far" is believing SiC + AI power permanently lift the earnings center of gravity; but the report cautions from the opposite side: the market may "see too far, mistaking an uncertain recovery for one already delivered." In 2025 GAAP operating margin was only 1.4%, and the current free-cash-flow yield is just a bit over 2%, below the roughly 4.56% of the 10-year Treasury — the premium the market pays for a distant recovery is precisely where the risk lies, not a cushion.

  Honest core judgment: contrary to most Baillie Gifford ideal names, onsemi's perception-gap direction is more likely "the market is overly optimistic" than "the market has not yet discovered it." The report's conclusion is incisive — this is the textbook case of a "good company, but bad price": business quality is above-average, but the current price is already near the upper edge of the optimistic-scenario intrinsic value (optimistic DCF $110-135), with an insufficient margin of safety.

  What would become the "narrative inflection point" (the sub-question added by the chained question) — look both ways:

  • The downward inflection (more to be wary of): revenue recovers for several consecutive quarters, but PSG utilization and operating margin still fail to rise → the market re-rates it from a "recovery growth stock" to an "ordinary cyclical," and the multiple contracts; or another large inventory/equipment impairment tied to strategic wavering (yet another after the 2025 $666.9 million and Q1 2026 $329.3 million); or continued large high-price buybacks proving to be a misallocation (the company already newly authorized $6 billion of buybacks in November 2025).
  • The upward inflection (the bull script): AI-data-center power + SiC truly ramping into a high-gross-margin pillar, with segment profit independently verifiable (resolving the "opaque disclosure" the report points out), turning "story-telling" into "a delivered lift in the earnings center of gravity."

  Baillie Gifford-lens verdict: the correct answer to this question for onsemi is "the market has not missed it — on the contrary, it is watching very hotly." It is not an undervalued dusty pearl but a highly cyclical stock fully priced on the recovery and AI narrative; the most realistic narrative inflection is more likely a downward re-rating of "recovery disproven → multiple reverting" than an upward explosion of "value discovery → Davis double-play." For a long-term owner, this means patiently waiting for a genuinely discounted price matters more than chasing an already-priced story right now.

  Jun 11, 2026